Manifestations

A manifestation is a grouped event — a product launch, partnership, patent cluster, funding round, and so on — not a single raw document and not a single graph node. Trendtracker builds manifestations from proof points and facts, then returns one row with supporting references.

This is the replacement for deprecated partnership and product-launch indicator endpoints.

Endpoint: POST /manifestations/search — playground

Types

Pass types to limit the mix. Omit it to search across types.

Investment and acquisition manifestations can attach both document proof points and knowledge-graph facts on the same row (references may mix proof_point and fact).

Typical Questions

• What product launches and partnerships happened around this topic last year?
• Which events are corroborated by more than one source?
• What investments and acquisitions sit next to the document-based events?

Paging

Manifestations use cursor paging, not skip:

• take — page size, max 100 (default 100)
• cursor — opaque token from the previous response (cursor is null when there is no next page)

Shared proof-point filters also apply: query, topicIds, organizationIds, languages, sourceIds, start / end.

Query

query uses the same Lucene-lite language as proof points, but only when types is restricted to proof-point types (news-event, patent, research-article, consulting-report, earnings-call, product-launch, partnership).

Fielded or boolean query (title:, AND / OR / NOT, grouping) is rejected when types is omitted or includes fact-backed types (organization, investment, acquisition). In those cases use a simple string: lithium, solid state battery, or "solid state battery".

Facts endpoints keep a plain organization-label query (max 50 characters). They do not accept this language.

Type-specific nested filters

Request Pattern

{
  "types": ["partnership", "product-launch"],
  "topicIds": ["11111111-1111-1111-1111-111111111111"],
  "start": "2023-01-01T00:00:00Z",
  "end": "2024-12-31T23:59:59Z",
  "take": 50,
  "partnershipFilter": {
    "minUniqueSourceCount": 2
  },
  "productLaunchFilter": {
    "minUniqueSourceCount": 2
  }
}

Follow-up page:

{
  "types": ["partnership", "product-launch"],
  "topicIds": ["11111111-1111-1111-1111-111111111111"],
  "cursor": "opaque-cursor-from-previous-response",
  "take": 50
}

Response Notes

Each manifestation can include:

• type
• title
• start / end
• companies, products
• amount
• uniqueSourceCount
• references — each has type (proof_point or fact), id, and either proofPointType or factType

The page object is { "data": [...], "cursor": "..." }.